Include a function to load multiple rasters on overlapping extent

[adehecq]

Let's say we want to compare a large raster and a small raster covering only a subsample of the larger raster. If the large raster is too large to fit in memory, or simply it takes long to load, it would be interesting to load the two rasters only over common area. Currently, doing this would require several lines of code. This is not really implemented if the two rasters have different CRS.

Same CRS:

r1 = gu.Raster(file1, load_data=False)
r2 = gu.Raster(file2, load_data=False)
intersection = r1.intersection(r2)
r1 = gu.Raster(file1, load_data=intersection)
r2 = gu.Raster(file2, load_data=intersection)

Different CRS:

r1 = gu.Raster(file1, load_data=False)
r2 = gu.Raster(file2, load_data=False)
intersection = gu.projtools.merge_bounds([r1.bounds, r2.get_bounds_projected(r1.crs)], "intersection")
r1 = gu.Raster(file1, load_data=intersection)
# Need to convert intersection into r2's CRS, which is not straightforward
r2 = gu.Raster(file2, load_data=intersection_r2)

Another option is to modify the behavior of Raster.reproject to not work on self.data, but on the dataset directly. This may become difficult after PR #265.

[adehecq]

Need to convert intersection into r2's CRS, which is not straightforward

Actually, this can be done in one line as in Raster.get_bounds_projected:
rio.warp.transform_bounds(src_crs, dst_crs, left, bottom, right, top, densify_pts)

[adehecq]

Closed via #307